Entropy as an event detector in a class of Hybrid Dynamic Systems. study case: A steam boiler

Abstract

In a Hybrid Dynamical System (HyDS) event occurrences can be generated by many causes. Also, these events may be controlled or uncontrolled (wanted or un-wanted) which in turn might generate imbalances and/or disorder in the system, with the potential of causing catastrophic faults. In order to be able to deal with controllable events regulation it is important solving an early detection problem, whenever there is no 100% reliable method for detecting them. The main objective of this work is to study and implement an innovative method to detect certain events in a HyDS, such as faults, starts and/or plant shutdowns, setpoint changes, among others. The proposed method is based on the estimation of entropy balance in the system and its efficiency as an event detector is illustrated using Matlab computer simulations on a steam boiler case of study. It is also showed how the entropy balance may be used as an event detector in different operation regions of the process.